There is a quiet irony in Tencent’s latest move. WeChat, the app that made the phrase “walled garden” synonymous with Chinese tech, is now unlocking its gates for AI agents. According to a report by Chinese financial news outlet Yicai, WeChat has been working with Huawei, Honor, Xiaomi, Oppo, and Vivo to allow their built-in voice assistants to send messages and initiate calls within the app.
Honor has already rolled out the feature on select devices using Yoyo, its AI-powered voice assistant. The move, reported on June 4 by South China Morning Post, marks a significant departure from Tencent’s historically guarded approach to third-party access.
But reading this purely as a Tencent story misses the bigger picture. The WeChat development is the clearest signal yet that AI agents are beginning to dismantle the logic that built super apps in the first place, and that signal carries weight well beyond China.
Super apps succeeded because they made leaving feel unnecessary. WeChat today accounts for 35% of mobile time in China, with Mini Programs reaching 960 million monthly users in 2025, transacting over US$500 billion annually. That stickiness was never accidental; it was engineered through friction. Every service you could access inside WeChat was one less reason to go elsewhere.
AI agents operate on exactly the opposite logic. The entire value proposition of an AI agent is that it removes friction. It navigates across apps on the user’s behalf, so the user never has to think about which platform holds which service. Instead of opening three different apps to plan an evening out, a single AI assistant can look up a restaurant, message contacts, and book a ride on its own.
That renders the walled garden obsolete. If an agent can step through the wall anyway, the wall stops being a moat. Tencent appears to recognise this. The Financial Times reported this week that Tencent is also developing its own internal AI agent for WeChat, one that would let users navigate millions of mini-programs through voice commands, with public testing possibly starting this month.
The dual strategy, opening the app to external agents and building an internal one, suggests Tencent is hedging rather than choosing.
The region should be paying close attention. Platforms like Grab and GoTo already serve hundreds of millions of users across e-commerce, ride-hailing, payments, and financial services, and their entire architecture is built on the same premise WeChat is now quietly abandoning: that being the one app users always return to is a defensible position.
Grab launched 13 AI-powered experiences at GrabX 2026 in April, positioning itself as an “intelligent everyday guide” for users across Southeast Asia. The move is smart, but it is also telling. Grab is embedding AI inside its own ecosystem, the same defensive manoeuvre Tencent is making. The question neither company has fully answered yet is what happens when a user’s AI agent of choice sits outside their platform entirely, and simply calls into it as one stop among many.
That is not a hypothetical. It is the direction the industry is heading. There are genuine complications ahead. Zhou Wei, head of Vivo’s Global AI Research Institute, flagged that enabling AI agents to operate across different apps raises serious security and authorisation challenges and that stricter standards will need to be built before this becomes seamless at scale.
Privacy concerns, authentication frameworks, and liability questions all remain unresolved. But those are engineering problems. The strategic question, whether super apps can remain the centre of gravity when AI agents start making the routing decisions, is harder to answer with a software update.
WeChat did not open its doors because it wanted to. It opened them because the alternative was becoming a locked room that AI simply walks around.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。